Best ways to test a new product line page before a full rebuild
Short answer: To test a new product line before a rebuild, publish one page on your current site, send a small amount of real traffic to it, and decide in advance what counts as a pass. A new product line page usually gets too few visitors for a proper A/B test (showing two versions to different visitors and comparing results), so you are testing whether anyone wants the offer, not which headline is better. Ploy, an AI platform that builds and hosts websites from a chat, is one way to draft the page. A document and a freelancer, a landing page tool, or doing it by hand can be cheaper.
You want to know if a new product line will sell before you spend months rebuilding the whole site around it.
I'm Ranjit, and I run Blur Studio: 15+ years building software, 120 apps taken live, and clients who have gone on to raise $10M+. I build AI systems and processes that help businesses do a lot more with a lot less time. I also work differently from most of what else is out there, more on that further down this page. Say hello.
Test the idea, not the website. A rebuild is expensive because it commits you to a structure. A single page tests the one thing you need to know first: when a real person sees this offer, will they act? You only need one page, one clear action, some real traffic and a decision rule written down before you start.
Below are the realistic ways to build the test page, what each costs as of November 2026, how to run the test without hurting your search results, and a worked example with made-up numbers. Ploy is covered as one option, with a clear list of when to skip it. If the offer is a small variant of what you already sell, the shorter guide on testing a new offer with a one-page site may be enough.
This page may contain affiliate links from which Blur Studio earns commissions.
What you are really testing
Pick one question before you pick a tool.
- Demand: do people want this product line at all? Measure enquiries, calls, sign-ups or quote requests.
- Price and offer: will they act at this price and with this promise? Change one thing at a time.
- Audience: which customers respond? Check where the enquiries come from.
Do not test the look of the site. A plain page that states the offer clearly tells you more than a polished page with a vague one.
Options from least effort to most
Prices are the vendors' own, shown in USD on their pages as of November 2026. Billing terms change the number, so check the vendor page.
| Option | What it is | Cost shown | Fits when |
|---|---|---|---|
| No tool: a document and a freelancer | You write the page in a document and a freelancer adds it to your site | Freelancer quote | You already have a site and a trusted person who can add a page fast. |
| Unbounce, a tool made only for landing pages | Build and host ad landing pages, with A/B testing on higher plans | Billed monthly: Starter $29, Build $99, Experiment $149, Optimize $249. Billed annually the page shows $22, $74, $112 and $187 a month. Agency starts at $349 a month, billed annually. 14-day trial, no card. | You run paid ads and want built-in split testing. A/B testing starts on the Experiment plan. |
| Lovable, an AI tool that builds web pages and apps from a chat | Describe the page, it generates it | Free $0 with 5 daily credits, up to 30 a month. Pro $25 a month, with a custom domain and 100 monthly credits. | You are happy to check generated code and publish on a separate address. |
| Do it by hand | Copy one of your own pages and edit it | Free apart from your time | Your platform lets you add a page easily and you only need one. |
| Ploy, one option of several | Describe the page in chat, review, publish | Free $0, Starter $50 a month, Pro $300 a month | You want a draft quickly in your brand and will review it yourself. |
If ads are the main way people reach the page, read best landing page builders for ads and ad landing pages that match the ad. If the page lives on your existing site, doing it by hand or with your freelancer is often the fastest and cheapest start.
Where Ploy fits
Ploy's AI website builder page says you point it at your existing site and it reads your content, design system and voice. You then describe a page in chat and its agent builds it with your brand, copy, images and a library of reusable page parts. You review and publish. The same page says you can ship one page without moving your whole site, while its publishing docs describe each publish as a whole-site snapshot, so ask how that works for your setup. Even so, it is a reasonable way to draft a test page in your own look.
Plan facts from Ploy's pricing page as of November 2026: Free is $0 with 500 AI credits a day up to 2,000 in total and a ploy.build address; Starter is $50 a month ($480 a year) with 4,000 credits a month and one custom domain; Pro is $300 a month ($2,880 a year) with 24,000 credits; Enterprise is by quote. Building draws credits, Ploy does not publish a cost per action, and credits do not roll over. A single page is a small job, so the Free plan can be enough to draft one, but you need a paid plan to put it on your own domain.
Ploy also has an experiments feature. Its docs say it tests one section of one page, with a champion (the current version) and a challenger (the new version) and a traffic split you choose from 10 to 90 percent. It must be switched on for your workspace on request, the site must be published with analytics on, and visitors who refuse cookies always see the original and are not counted. The docs do not describe the statistics method or a significance threshold, so ask before you rely on its verdict. See Ploy vs Unbounce if split testing is your main need.
Whoever drafts the page, AI-written pages need a human check. Read every claim, price and photo before it goes live.
Cheap checks to run before you build any page
- Ask. Message five past customers: "I am thinking of offering X. Would you want it, and what would you pay?" Real answers beat guesses.
- Email your list. One short email with the offer and a reply button. Count replies.
- Check search demand. Type the product into Google and see who already sells it and how they describe it.
- Take names, not payment. A form that says "tell me when this is ready" is a weaker signal than a quote request, so set a higher pass line for it.
Where to put the page
- On your main site in a new folder, such as yoursite.com/new-product. It sits with your existing pages. Google's John Mueller, a Google search spokesperson, said in 2018 that Google generally treats subdomains and subfolders the same and he would personally keep things together, using subdomains for things that are really different.
- On a subdomain, such as new.yoursite.com. Easier if your main site is hard to change.
- On a separate address from a tool. Quick, but it does not carry your brand and you may need to redirect later.
Ploy can act as a reverse proxy (a server in front of your site that passes each request to the right place). Its path rules can send one path to a Ploy site and everything else to your old host, which lets a new page live under your own domain without moving the rest. Members can manage domains and path rules. Ploy's docs say rule changes apply at the edge in seconds. If the save shows a sync error, do not assume it is live. Check the page itself.
Do not hurt your search results while testing
- Do not show Google one page and people another. Google calls that cloaking.
- If a test sends people to a different address, use a 302 (temporary) redirect, not a 301 (permanent). Google's website testing guide also says to use rel="canonical" (a tag that names the preferred version) on the alternative, pointing to the original.
- Run tests only as long as necessary and remove test pieces promptly.
- Keep unfinished drafts out of search. Ploy's docs recommend a noindex tag (a note that tells Google not to list the page) on staging pages published to Preview only. A noindex tag only works if robots.txt does not block the page.
- Do not make near-identical pages for many places or products. Google's spam policy calls pages made to rank for similar queries doorway abuse. One page for one offer is fine.
How much traffic you need, and what to do when you have little
Evan Miller's sample size calculator gives a feel for the numbers. Even a modest lift usually needs thousands of visitors for each version. A new product line page rarely gets that. His article on running tests also warns about stopping early: in his worst-case example, checking after every visitor and stopping at the first significant result gave a false win 26.1 percent of the time, against the 5 percent you would expect. Optimizely's guide says to run a test for a minimum of one business cycle (seven days).
So for a small test do not split versions. Use a simple pass and fail on one version:
- Decide the action: call, form or booking.
- Decide the traffic: for example 150 visitors from your emails, social posts or a small ad budget (made-up numbers).
- Decide the pass: for example at least 6 enquiries. Decide the fail: 2 or fewer. In between, change the offer and run it again once.
- Write all three numbers down before you publish.
When traffic is larger, see how to test two versions of a landing page.
Worked example (made-up numbers)
A garden furniture seller wants to add a made-to-order outdoor kitchen range. A rebuild would cost months. Instead:
- Day 1: pass line "at least 6 quote requests from about 300 visitors", which is 2 percent. Page text written in a document: who it is for, three options with from-prices, delivery time, one quote form.
- Day 2: page built and checked. Phone and form tested. noindex on the draft until review is done.
- Days 3 to 16: an email to past customers and a small ad budget send 300 visitors.
- Result: 11 quote requests from 300 visitors (3.7 percent), 5 of them from the email list. The pass line was 6. The owner moves to a second test on price, and delays the rebuild until two rounds agree.
These numbers are invented to show the method. Your pass line depends on your margin and your cost per enquiry.
Copy-ready test checklist
- One question, one page, one action.
- Pass and fail numbers written down before publishing.
- Phone number and form tested on a phone.
- Analytics events firing (page view and the action).
- Page consistent with any ad or email that sends people to it.
- Draft kept out of search until reviewed.
- Review date in your calendar.
- Plan for the page after the test: keep, improve, redirect or remove.
Who should skip this, and who should skip Ploy
- If you only need one simple page, a document link or a free builder is enough. Skip any paid tool.
- If you need heavy paid-ad testing with many variants, a landing page tool such as Unbounce is built for that. Ploy's experiments run on request and test one section of one page.
- If your product line needs live stock, a basket or bookings, that is not a test page. It needs the real system.
- If you want a design a template cannot match, or an app or portal, ask a specialist. Blur Studio builds custom work.
How I'd approach this
I would not rebuild anything until two rounds of one cheap page say the demand is real. I would write the offer in one sentence, set the pass line, and build the page the cheapest way my site allows. If I wanted a quick branded draft, I would try Ploy's free plan for the draft and check every line myself. I would use a paid plan only to put the page on my own domain, or switch to a landing page tool if my main job were paid ads with several variants.
First-week plan. Day 1: one sentence for the offer and the pass line. Day 2: write the page text. Day 3: build, check on a phone, test the form. Day 4: send it to your list and start any small ad spend. Days 5 to 7: do not edit, just count. Check again after one to two weeks. If it passed, plan the real build. If it failed, change one thing and run once more, or drop the idea cheaply.
You may also want previewing a site rebuild before you commit.
Related guide: adding landing pages to an existing website without moving your domain.
Questions people ask
How long should I run a landing page test?
Optimizely's guide says a minimum of one business cycle, which it puts at seven days. Decide the length and the traffic before you start and do not stop early because the numbers look good.
Can I test a page without hurting my Google rankings?
Yes, if you follow Google's testing guide: no cloaking, use 302 redirects for temporary tests, canonical tags on alternatives, and remove test pieces when done. Keep unfinished drafts out of search with a noindex tag. See how to show a draft without Google finding it.
Do I need A/B testing for a new product line?
Probably not at first. Detecting a modest lift usually needs thousands of visitors per version, which a new page rarely gets. Use one version with a written pass and fail rule, then compare versions later when traffic grows.
How much does it cost to test a page with Ploy?
As of November 2026 Ploy shows Free at $0, Starter at $50 a month and Pro at $300 a month. Drafting one page can fit in the Free plan's credits, but your own domain needs a paid plan. Credit cost per action is not published, so check the pricing page. See Ploy pricing explained.
Should the test page go on my main site or a new address?
A folder on your main site keeps the page with your existing pages. John Mueller of Google said in 2018 that Google generally treats subdomains and subfolders the same, and he would personally keep things together. A subdomain is easier if your main site is hard to change.
What if my test fails?
That is a result. Change one thing, such as the price or the headline promise, and run once more with the same pass line. If it fails twice, drop or reshape the idea. You have saved the cost of a rebuild.
Can I use Lovable or Unbounce instead of Ploy?
Yes. Unbounce is built for landing pages and ads, with A/B testing from its Experiment plan. Lovable builds pages from chat prompts and publishes on its own address on the free plan. See Ploy vs Lovable and Ploy vs Unbounce.
More in this topic: Websites: launching, refreshing and previewing a site (7 guides) →
Sources
- Ploy pricing page (Ploy)
- Experiments (Ploy docs)
- Unbounce pricing (Unbounce)
- Lovable pricing (Lovable)
- Website testing and Google Search (Google Search Central)
- Block search indexing with noindex (Google Search Central)
- Spam policies for Google web search (Google Search Central)
- How long to run an experiment (Optimizely)
- A/B test sample size calculator (Evan Miller)
- How Not To Run an A/B Test (Evan Miller)
- Google treats subdomains and subdirectories (Search Engine Journal, 2018)